Working at a software company means that I’m more than familiar with phrases like “advanced technology” and “breakthrough algorithms”… sometimes too familiar! Although these are actually decent descriptions, sometimes it’s nice to know exactly how the software works in order to be able to use it most effectively, instead of just knowing that it’s “advanced”.

Now, this is immediately relevant because we just released Topaz Detail, a plug-in that specializes in detail enhancement without halos. It does use pretty new technology – for example, a detail enhancement product that doesn’t create halos would have been impossible to develop just a year ago.

So immediately people emailed us wanting to know how it worked. One of the more common questions was why we implemented a pre-processing technique only in Topaz Detail rather than our other software. Well, the answer is so simple that it can be related in haiku form:

Internal splitting
Precise size-based enhancement
Topaz Detail rocks!

I know that probably answered all your questions. Here’s a more complex answer anyways, manifested in a delightful blue diagram:

Topaz Detail internal flow. Click for larger. (You may have to scroll right)
Topaz Detail flow. Click for larger. (You may have to scroll right)

Click on the diagram for a larger and more voluptuous version. I’ll walk you through the internal steps that Detail takes from start to finish.

1. First, you have your original image. You now access Topaz Detail and appreciate the fine aesthetics of the pre-processing screen.

2. Pre-processing starts. Topaz Detail separates your image into one based on chrominance (color information) and one based on luminance (grayscale information). These will undergo different processing methods and will be affected by different sliders.

3. The luminance information is further internally broken down into three detail layers based on size, and a base layer. Manipulating these four layers make up the bulk of the detail enhancement functionality of the software. After this step, pre-processing ends and the Topaz Detail user interface pops up. (This is also the portion of pre-processing that runs the slowest… probably because it has the most advanced breakthrough technological algorithms.)

4. The user adjusts the large, medium, and small detail sizes, and each individual detail layer is affected appropriately. If you like, you can see what each of the four individual layers looks like. To see the base layer by itself, turn the large, medium, and small sliders to 0. To see each individual detail size layer, drag that particular one all the way to the right and set the contrast to 0 (which will switch the base layer to neutral gray).

5. The luminance layer is re-combined with the adjusted small, medium, and large detail layers as well as the base layer.

6. Highlight and shadow protection, as set by the user, is applied to the luma layer.

7. The internal luma and chroma layers are re-combined to form your startlingly good final processed image.

Steps 5, 6, and 7 take no time at all. This means that, although there can be a sometimes annoying wait for the initial pre-processing stage, there will be no actual waiting after that. All adjustments are made instantaneously after the necessary initial pre-processing.

This is the complete internal workflow that Detail uses. Now that you know, it becomes quite obvious how it can sharpen and enhance detail without creating any halos – simply because it doesn’t really “sharpen”! All it does is increase the intensity of the appropriate detail layer, which gives a sharpening effect without any of the negative artifacts normally associated with sharpening or detail enhancement.

This post was originally posted on August 18, 2009
Share this
Share on Facebook0Tweet about this on TwitterShare on Google+0Pin on Pinterest0Share on LinkedIn0Share on Reddit0Share on Tumblr0Email this to someonePrint this page
Tagged , , , , .